Dear colleagues,
The winner of the 2025 ICS Student Paper Award is Hao Hao (Carnegie Mellon). You can find below the award committee citation:
- Title: "Robust Paths: Geometry and Computation"
- Authors: Hao Hao and Peter Zhang
- Student: Hao Hao
- Citation: The authors study the problem of approximating the set of optimal solutions to a robust optimization problem as the uncertainty set size varies. This is a challenging problem where the optimal radius is usually selected via cross-validation, or by leveraging distributional information to invoke potentially conservative probabilistic guarantees. To overcome this challenge, the authors characterize sets of optimal solutions, or "robust paths", as projections of Bregman curves onto the feasible set. They further demonstrate that these robust paths are well-approximated by the trajectories of deterministic optimization algorithms, potentially allowing ideas from robust optimization to be applied in settings where robust optimization is intractable. The paper is well-written and concludes with several computational studies, making it a deserving recipient of the award.
The award committee also elected Thomas Hübner (ETH Zürich) as a runner-up. Citation below:
- Title: "Spatial branch-and-bound for nonconvex separable piecewise linear optimization"
- Authors: Thomas Hübner, Akshay Gupte, and Steffen Rebennack
- Student: Thomas Hübner
- Citation: The authors propose a novel spatial branch-and-bound algorithm for solving the separable nonconvex piecewise linear optimization problem based on convex envelope techniques. They demonstrate convergence under mild assumptions and show significant computational improvements over existing mixed-integer linear programming formulations.
Finally, the award committee also selected, in no particular order, Yupeng Wu (London Business School) and Matías Villagra (Columbia) as honorable mentions. Citations below:
- Title: "The Surprising Performance of Random Partial Benders Decomposition"
- Authors: Yupeng Wu and Jean Pauphilet
- Student: Yupeng Wu
- Citation: This paper introduces a simple yet universally applicable variant of Benders decomposition that randomly retains a subset of continuous variables in the master problem. Computational experiments demonstrate that this random retention strategy can perform as effectively as problem-specific approaches, highlighting its potential to advance the current computational practice.
- Title: "Accurate Linear Cutting-Plane Relaxations for ACOPF"
- Authors: Daniel Bienstock and Matías Villagra
- Student: Matías Villagra
- Citation: For the ACOPF problem, the authors obtain strong, numerically stable bounds using a cutting plane approach, aided by reformulations and proper cut management. The authors effectively recycle past cuts to warm-start the optimization of related instances, including for the multi-period case.
I would like to thank the award committee for their work to identify these outstanding papers from a pool with so many other great pieces of scholarship. This committee was chaired by Austin Buchanan (Oklahoma State University), and also included Ryan Cory-Wright (Imperial College London), Yongchun Li (University of Tennessee), and Young Woong Park (Iowa State University).
Please join us at the ICS business meeting during the INFORMS 2025 Annual Meeting in Atlanta on Monday, October 27, to recognize the work done by our colleagues as part of their PhD and to thank the award committee.
Sincerely,
------------------------------
Thiago Serra
Assistant Professor of Business Analytics, University of Iowa
INFORMS Computing Society Chair (2024-2025)
------------------------------